Regulation of Protein-tyrosine Phosphatases (cid:1) and (cid:2) by Calpain-mediated Proteolytic Cleavage*

The precise subcellular localization of non-receptor tyrosine phosphatases is a major factor in regulating their physiological functions. We have previously shown that cellular processing of protein-tyrosine phosphatase (cid:2) (PTP (cid:2) ) generates a physiologically distinct, cytoplasmic form of this protein, p65 PTP (cid:2) . Here we describe a novel protein form of the related receptor-type tyrosine phosphatase (cid:1) (RPTP (cid:1) ), p66 PTP (cid:1) , which is detected in nearly all cell types where RPTP (cid:1) is expressed. Both p66 PTP (cid:1) and p65 PTP (cid:2) are produced by calpain-mediated proteolytic cleavage in vivo . Cleavage is inhibited in living cells by a variety of calpain inhibitors, can be induced in primary cortical neurons treated with calcium chloride, and is observed in lysates of brain or of cultured cells following addition of purified calpain. Cleavage occurs within the intracellular juxtamembrane domain of RPTP (cid:1) , releasing the phosphatase catalytic domains from their membranal anchors and translocating them to the cytoplasm. Translocation reduces the ability of PTP (cid:1) to act on membrane-associated substrates, as it loses its ability to dephosphorylate Src at its C-terminal regulatory site, and its ability to dephosphorylate the Kv2.1 voltage-gated

sion of RPTP␣ has also been associated with low tumor grade in human breast cancer (37) and with late stage colon carcinomas (38). RPTP␣ is believed to undergo inhibitory dimerization, in the course of which the helix-turn-helix wedge domain of each RPTP␣ molecule interacts with and blocks access to the active site of its dimerization partner (39 -42). No ligand for RPTP␣ is known, although the glycosylphosphatidylinositol-linked protein contactin has been shown to form a complex with RPTP␣ in neuronal cells (43), and recent studies have suggested that newborn calf serum may contain ligands for the phosphatase (44).
The existence of non-membranal forms of PTP⑀ prompted us to search for similar forms of RPTP␣, which might affect the nature of PTP␣ activity in cells. We report here that RPTP␣ can be processed in vivo to generate p66 PTP␣, an N-terminally truncated form analogous to p65 PTP⑀. Both p66 PTP␣ and p65 PTP⑀ are produced in vivo from larger RPTP␣ or PTP⑀ molecules by calpain-mediated cleavage. As it lacks membraneanchoring domains, p66 PTP␣ is inherently a cytoplasmic molecule, although it can be detected in part at the cell membrane when expressed together with full-length RPTP␣ molecules. When absent from the cell membrane, p66 PTP␣ is incapable of dephosphorylating Src at its C-terminal regulatory site and has significantly reduced activity toward the Kv2.1 potassium channel. These results demonstrate the importance of membrane association for the known functions of RPTP␣, and underscore major functional differences between p66 PTP␣ and its full-length RPTP␣ precursor.
Cell Culture-Human embryonic kidney 293 cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.), supplemented with 10% (v/v) fetal calf serum (Life Technologies, Inc.), 2 mM glutamine, 50 units/ml penicillin, and 50 g/ml streptomycin. Transformed mouse fibroblasts deficient for the Src, Yes, and Fyn kinases (SYF cells; Ref. 47) were grown in similar medium containing 4 mM glutamine and 1 mM sodium pyruvate. 293 cells were transfected by the calcium phosphate method (48); SYF cells were transfected with Lipo-fectAMINE 2000 (Life Technologies, Inc.) according to the manufacturer's instructions. In some experiments cells were treated for 2 h with 50 M MG132, 5-50 M calpeptin, 100 M leupeptin, or vehicle (up to 1.25% Me 2 SO). Fractionation of cells into membranal and cytosolic fractions by mechanical disruption and centrifugation has been described (18). Primary cortical neurons were prepared by the trituration method as described (49). Following trituration and washing, neurons were resuspended in phosphate-buffered saline, 0.6% glucose, and aliquots of 1 ϫ 10 6 cells were incubated as such or exposed to 1 mM CaCl 2 for the times indicated in Fig. 4. For calpain inhibition, neurons were incubated for 5 min with 20 M calpeptin prior to addition of calcium chloride. Neurons were then pelleted, frozen in liquid nitrogen, and extracted in RIPA buffer containing 10 g/ml each of aprotinin, PMSF, and leupeptin, and 1 g/ml pepstatin. Cell lysates were analyzed by protein blotting as described below.
Immunoprecipitation and Protein Blot Analyses-Unless noted otherwise, cells or mouse organs were lysed in buffer A (50 mM Tris-Cl, pH 7.5, 100 mM NaCl, 1% Nonidet P-40), supplemented with 0.5 mM sodium pervanadate and protease inhibitors (1 mM N-(␣-aminoethyl)benzenesulfonyl fluoride, 40 M bestatin, 15 M E64, 20 M leupeptin, 15 M pepstatin; Sigma). In several preliminary experiments, the lysis buffer also contained 5 mM EGTA, although this did not affect the pattern of proteins detected and was therefore not used in later experiments. 5-20 g of total protein were analyzed on 7% SDS-polyacrylamide gels, followed by transfer to nitrocellulose membranes (Protran, Schleicher & Schuell), and hybridization to antibodies. Complete protein transfer following blotting was verified routinely by noting transfer of prestained molecular size marker proteins of the proper size range; absence of lane-to-lane variations in blotting was verified by staining the blotted membranes with Ponceau S (Sigma). For immunoprecipitations, 0.5-1 mg of total cell protein were incubated with anti-FLAG M2 affinity beads (Sigma) or with anti-phosphotyrosine antibodies and protein A beads (Amersham Pharmacia Biotech) for 3-4 h, followed by three extensive washes with RIPA buffer. Experiments were repeated two to five times, and representative blots are shown.
Pulse-chase analysis was performed in 293 cells 24 h following transfection with a FLAG-tagged cyt-PTP⑀ cDNA. Cells were washed in serum-free Dulbecco's modified Eagle's medium lacking methionine and cysteine (Sigma) and were then labeled in 4 ml of the same medium supplemented with 48 l (ϭ 0.48 mCi) of [ 35 S]methionine (1000 Ci/ mmol, 10 mCi/ml, Amersham Pharmacia Biotech) for 30 min. Following removal of the radioactive medium, cells were washed twice in phosphate-buffered saline and incubated for up to 8 h in serum-containing growth medium supplemented with 2 mM each of non-radioactive methionine and cysteine. At selected time points cells were lysed, immunoprecipitated with anti-FLAG beads, and blotted. Radioactivity present in each band of PTP⑀ was quantified using a phosphorimager (BAS 2500, Fuji, Japan); the same blots were then probed with anti-PTP⑀ serum and scanned with a scanning densitometer to correct for differences in PTP⑀ protein amounts between lanes.

RESULTS
p66 Is an N-terminal Truncated Form of RPTP␣-Anecdotal examination of RPTP␣ expression in various cell lines and mouse tissues consistently revealed a protein of ϳ66 kDa in addition to the previously described full-length unglycosylated (ϳ100 kDa) and glycosylated (ϳ130 kDa) forms of RPTP␣ ( Fig.  1; Ref. 26). This protein, which we refer to as p66 PTP␣, was clearly PTP␣-derived, as it appeared together with full-length RPTP␣ in 293 cells transfected with a complete PTP␣ cDNA expression construct (Fig. 1A). Furthermore, p66 PTP␣ and RPTP␣ were both detected in protein lysates from brains of wild-type or of PTP⑀-deficient mice, but were both absent from similar lysates prepared from brains of PTP␣-deficient mice ( Fig. 1B; Ref. 31). The expression patterns, expression levels, and size of p66 PTP␣ strongly resembled those of the p67 and p65 forms of the closely related PTP⑀ (Ref. 16; Fig. 1). These forms of PTP⑀ lack N-terminal sequences of full-length PTP⑀, suggesting that p66 PTP␣ might be an analogous N-terminal truncated form of RPTP␣. In agreement with this, addition of a FLAG tag at the 3Ј end of the coding region of the PTP␣ cDNA resulted in appearance of the tag in both p66 PTP␣ and RPTP␣ proteins (Fig. 1A), indicating that the C terminus of p66 PTP␣ was intact.
p66 PTP␣ Is Produced by Proteolytic Cleavage of Full-length RPTP␣ Molecules-The p67 form of PTP⑀ is produced by initiation of translation at an internal ATG codon corresponding to Met85 of tm-PTP⑀, whereas p65 PTP⑀ is produced by proteolytic processing of tm-PTP⑀, cyt-PTP⑀, or p67 PTP⑀ (16). Sitedirected mutagenesis studies have shown that cleavage is independent of, but occurs in the immediate vicinity of, Met-99 of tm-PTP⑀ ( Fig. 2; Ref. 16). Examination of the juxtamembrane sequences of RPTP␣ revealed that neither Met-85 nor Met-99 of tm-PTP⑀ were conserved in RPTP␣ and that this region contained no other close ATG codons, reducing the likelihood that p66 PTP␣ was the product of internal initiation of translation. However, sequences surrounding Met-99 in tm-PTP⑀ were highly conserved in the RPTP␣ protein (Fig. 2), raising the possibility that the yet-unidentified cleavage site of PTP⑀ is conserved in RPTP␣. In a manner consistent with this, the proteasome and protease inhibitor MG132 reduced amounts of p66 PTP␣ detected in cells (Fig. 3A), similar to its ability to inhibit production of p65 PTP⑀ from full-length PTP⑀ (16). Additional studies, which we detail below, further strengthen the conclusion that p66 PTP␣ is produced by proteolytic cleavage of larger PTP␣ molecules and is therefore analogous to p65 PTP⑀. Of note, lysis of cells and tissues was routinely performed in the cold and in buffers containing inhibitors of proteases, including calpain inhibitors as explained below. p65 PTP⑀ and p66 PTP␣ were observed also in protein blots prepared from cells that had been lysed directly in boiling SDS-PAGE loading buffer (data not shown), indicating that these shorter proteins were present in cells prior to lysis.
p66 PTP␣ and p65 PTP⑀ Are Products of Calpain-mediated Processing in Vivo-MG132, which inhibits accumulation of p65 PTP⑀ and p66 PTP␣, is a known inhibitor of the proteasome and of several Ca 2ϩ -regulated and lysosomal proteases (50,51). Previous studies have shown that lactacystin and epoxomicin, which are more specific inhibitors of proteasome function than MG132, did not affect accumulation of p65 PTP⑀ (16). This finding indicated that MG132 affects p65 PTP⑀ production not by inhibiting proteasome function, but by inhibiting another protease(s). As MG132 can also inhibit the major calcium-regulated cysteine protease calpain, we examined whether p65 PTP⑀ and p66 PTP␣ are products of calpainmediated cleavage. For this purpose we co-expressed PTP⑀ or RPTP␣ in cells together with the highly specific calpain inhibitory protein calpastatin. Expression of calpastatin significantly reduced production of p65 PTP⑀ from cyt-PTP⑀ or tm-PTP⑀, and p66 PTP␣ from RPTP␣ ( Fig. 3B and results not shown). Similar results were obtained when cells expressing cyt-PTP⑀ were treated with the cell-permeable calpain inhibitor calpeptin, with significant inhibition of p65 PTP⑀ accumulation evident in cells treated with as little as 5 M calpeptin (Fig. 3C). Calpeptin also inhibited formation of p66 PTP␣ and p65 PTP⑀ from RPTP␣ or tm-PTP⑀, respectively ( Fig. 3A and results not shown). The cell-permeable cysteine protease inhibitors leupeptin and E64d also significantly reduced the amounts of p65 PTP⑀ in cells, whereas pepstatin, PMSF, chloroquine, and ammonium chloride, inhibitors known to act on other proteases and proteolytic systems, had no effect (Fig. 3D). In a separate set of experiments, leupeptin inhibited accumulation of endogenous p65 PTP⑀ in NIH3T3 cells (data not shown).
Calpain-mediated cleavage of endogenous RPTP␣ and tm-PTP⑀ could be induced in primary cortical neurons prepared from day 14.5 mouse embryos. These neurons express high levels of RPTP␣ and very low levels of tm-PTP⑀ (Fig. 4). 2 p66 PTP␣ was not detected in freshly isolated neurons prior to treatment, but was readily visible following addition of 1 mM CaCl 2 to the cells. Cleavage was not detected in neurons, which had been exposed to calpeptin prior to CaCl 2 treatment, attesting to the involvement of calpain in the cleavage event ( Fig. 4).
In similar experiments the cell-impermeable calpain inhibitor E64 did not inhibit calcium-dependent cleavage, indicating that the cleavage observed occurred in intact cells. Interestingly, induction of cleavage by CaCl 2 did not result in processing of all RPTP␣ molecules present in the primary neurons. This result did not change when the stimulus applied to the cells included 200 M glutamate (not shown), suggesting that cleavage is regulated by additional cellular mechanisms as is discussed below. Taken together, these data indicate that p65 PTP⑀ and p66 PTP␣ are produced in vivo by calpain-mediated cleavage of larger PTP⑀ or RPTP␣ molecules, respectively.
Calpain Cleaves PTP␣ and PTP⑀ in Vitro-In a manner consistent with the above results, addition of purified calpain to lysates of cells expressing cyt-PTP⑀ resulted in cleavage of cyt-PTP⑀ and increased amounts of a protein that co-migrated with p65 PTP⑀ in SDS-PAGE gels (Fig. 5, A and B). Cleavage was clearly due to the added calpain, with the extent of cleavage dependent upon the amount of calpain added (Fig. 5A). Similar results were obtained when calpain was added to lysates of cells expressing tm-PTP⑀ or RPTP␣, or to extracts of mouse brain containing endogenous tm-PTP⑀ and RPTP␣ (Fig.  5B). Cleavage of RPTP␣ and PTP⑀ by calpain was specific and did not result in total degradation of these proteins. Addition of calpain to the protein extracts shown in Fig. 5B did not result in widespread, nonspecific cleavage of other cellular proteins, as judged by Coomassie Blue staining of SDS-PAGE gels prepared from these extracts. However, specific, known substrates of calpain, such as tubulin, were also cleaved following calpain treatment of these lysates (data not shown).
p65 PTP⑀ Is a Stable Protein-Amounts of p65 PTP⑀ and p66 PTP␣ are significantly decreased as early as 2 h after addition of MG132, calpeptin, or leupeptin, indicating that these processed proteins are either inherently short-lived or are destabilized following general inhibition of calpain. In order to compare the stability of the various forms of PTP⑀, we performed a series of pulse-chase experiments in which cyt-PTP⑀, p67, and p65 proteins were expressed in 293 cells from the cyt-PTP⑀ cDNA. Following labeling with [ 35  blotted. Radioactivity present in each of the PTP⑀ bands was quantified with the aid of a phosphorimager, after which the same protein blots were probed with anti-PTP⑀ antibodies to allow comparison of the measured radioactivity with the amounts of PTP⑀ protein present in each band.
As seen in Fig. 6, full-length cyt-PTP⑀, p67 PTP⑀, and p65 PTP⑀ were stable to similar extents, with slight decreases detected in the normalized amount of radioactivity associated with them during the chase period shown. Interestingly, treatment of cells with leupeptin significantly reduced the amount of p65 PTP⑀ protein present at the 2-h time point by 80%, together with a similar 67.3% reduction in the amount of radioactivity present in this band ( Fig. 6B and results not  shown). Consequently, leupeptin treatment did not change significantly the specific radioactivity associated with p65 PTP⑀ (Fig. 6A), although amounts of p65 protein were significantly reduced (Fig. 6B). This last result indicates that a consequence of inhibition of calpain by leupeptin is destabilization of p65 PTP⑀, but that in the absence of calpain inhibition p65 PTP⑀ is stable. These findings suggest that calpain also plays a role in stabilizing p65 PTP⑀ once the protein has been produced. The demonstrated ability of exogenous and endogenous calpain to cleave PTP⑀ and PTP␣ (Figs. 4 and 5) provides evidence, which is independent of calpain inhibitors, that calpain participates also in cleaving both PTPases. Calpain then plays a dual role in production of p65 PTP⑀, both in the actual cleavage event and in stabilization of the resulting cleavage product. This effect of calpain inhibitors is limited to p65 PTP⑀ (and presumably to p66 PTP␣ as well), as amounts of the full-length forms of PTP⑀ and RPTP␣ and of p67 PTP⑀, which are known to be produced by mechanisms not involving proteolysis, were not reduced in the presence of leupeptin or of other calpain inhibitors ( Fig. 3; Ref. 16).
Proteolytic Processing Translocates PTP␣ to the Cytoplasm-Full-length RPTP␣ is an integral membrane protein. Processing of RPTP␣ to yield p66 PTP␣ occurs downstream of the transmembranal domain of RPTP␣; hence, p66 PTP␣ lacks this domain and should be a cytoplasmic protein. In order to examine this possibility, we performed a series of subcellular fractionation experiments in cells expressing PTP␣ (Fig. 7). The precise N terminus of p66 PTP␣ has yet to be determined, but comparison with p65 PTP⑀ suggests that p66 PTP␣ should begin at or in the immediate vicinity of Leu-198 (Fig. 2). We therefore constructed a truncated PTP␣ cDNA molecule, in which the initiating ATG codon was inserted immediately upstream of Leu-198; the protein produced from this cDNA construct co-migrated with p66 PTP␣. Fractionation of cells expressing this truncated form of PTP␣ revealed that it was in fact entirely cytoplasmic (Fig. 7). In contrast, fractionation of cells expressing full-length PTP␣ cDNA and in which both RPTP␣ and p66 PTP␣ proteins were present revealed that RPTP␣ was present exclusively in the membrane fraction, while p66 PTP␣ was detected in both the cytoplasmic and membranal fractions (Fig. 7). Retention of p66 PTP␣ in the membrane fraction in the presence of RPTP␣ could be accounted for by the existence of PTP␣ dimers (40), with the cleaved product of one RPTP␣ molecule being associated with another uncleaved, membrane-bound RPTP␣ molecule.
Significantly Reduced Activity of p66 PTP␣ toward Src and the Kv2.1 Potassium Channel-Translocation of the catalytic domains of PTP⑀ and PTP␣ to the cytosol could serve a regulatory function by either reducing the activity of these PTPases toward membrane-associated substrates, or by increasing their activity toward substrates located in the cytosol. Very little is known about the cytosolic functions of both PTPases; hence, we chose to focus in this study on the consequences of cleavage on membrane-associated substrates. One of the best documented physiological functions of RPTP␣ is to dephosphorylate the membrane-associated kinase Src at its C-terminal tyrosine residue (Tyr-529 in mouse), thereby activating the kinase and initiating a broad series of cellular events (30,31). In order to examine the ability of p66 PTP␣ to act on Src, mouse fibroblasts lacking Src, Yes, and Fyn (SYF cells; Ref. 47) were transfected with expression vectors for murine wild-type Src and for either RPTP␣ or the cytoplasmic truncated form of PTP␣ described above. Absence of endogenous Src and very low levels of endogenous PTP␣ in SYF cells ensured that Src molecules examined were only those present in cells co-transfected with PTP␣, and that the effect of PTP␣ on Src would not be masked by endogenous Src from untransfected cells. Phosphorylation of Src at Tyr-529 was examined by probing protein blots prepared from the relevant cell extracts with an antibody specific for pY529Src. Co-expression of Src together with fulllength RPTP␣ resulted in a 43% reduction in phosphorylation of Src at Tyr-529, in agreement with results obtained previously by several groups (Fig. 8; Ref. 52). In contrast, co-expression of truncated PTP␣ did not significantly affect Src phosphorylation at Tyr-529, despite higher expression levels of this form as compared with RPTP␣ (Fig. 8, lane 3 versus lanes 4 and  5). As indicated under "Experimental Procedures", care was taken to ensure that protein blotting was complete and that increased levels of p66 PTP␣ versus RPTP␣ proteins faithfully reflect the situation within the cells.
We have documented previously that cyt-PTP⑀ dephosphorylates and inactivates the delayed rectifier, voltage-gated potassium channel Kv2.1, in a manner that correlates with reduced myelination of peripheral nerve axons in PTP⑀-deficient mice. In this system, PTP⑀ and Src appear to counter each other's activity toward their common substrate, Kv2.1 (25). In order to examine the possibility that RPTP␣ could also dephosphorylate Kv2.1, we transfected 293 cells with Kv2.1 and constitutively active (Y527F) chicken Src, either in the presence or absence of RPTP␣ (Fig. 9). Activated Src was used in this study to ensure phosphorylation of Kv2.1, as well as to prevent PTP␣ affecting Src activity via dephosphorylation of Tyr-527. Indeed, co-expression of Kv2.1 with Src resulted in massive tyrosine phosphorylation of the channel protein, whereas presence of RPTP␣ in these cells reduced Kv2.1 phosphorylation by 97% (Fig. 9).
Kv2.1 is an integral membrane protein. In order to determine whether the non-membranal localization of p66 PTP␣ impedes its ability to act upon Kv2.1, we examined the ability of p66 PTP␣ to reduce phosphorylation of Kv2.1 in transfected 293 cells. Although RPTP␣ nearly eliminated Kv2.1 phosphorylation, expression of significantly higher levels of p66 PTP␣ reduced Kv2.1 phosphorylation by only 32% (Fig. 9, lane 5). Strong dephosphorylation of Kv2.1 was observed only following massive overexpression of p66 PTP␣, which resulted in severalfold more p66 PTP␣ protein being present in comparison with RPTP␣ (Fig. 9, lane 3). These results indicate that membrane localization is a central determinant of the ability of RPTP␣ to act upon Kv2.1, although it appears not to be the only factor regulating this activity. Together with the inability of p66 PTP␣ to act upon Tyr-529 of Src, these results indicate that translocation of PTP␣ to the cytoplasm significantly reduces its ability to act upon molecules located at the cell membrane and underscores an important functional difference between p66 PTP␣ and RPTP␣. DISCUSSION Data presented in this study indicate that p66 PTP␣ and the analogous p65 PTP⑀ are produced from larger RPTP␣ or PTP⑀ molecules by calpain-mediated proteolytic processing. Accumulation of both molecules can be prevented in vivo by the same series of calpain inhibitors: calpastatin, calpeptin, MG132, and leupeptin. Furthermore, cleavage of RPTP␣ to yield p66 PTP␣ can be induced in primary cortical neurons by the presence of calcium cations, and this process can be prevented by prior inhibition of calpain in these cells. Exogenous calpain cleaves RPTP␣ and PTP⑀ in a specific manner to yield proteins that co-migrate with p66 PTP␣ and p65 PTP⑀, respectively. Pulsechase experiments indicate that p65 PTP⑀ and most likely p66 PTP␣ are stable proteins, but are destabilized in the course of generalized calpain inhibition. The data then suggest that calpain activity is required for both production and subsequent stabilization of p65 PTP⑀ and p66 PTP␣.
Interestingly, calcium-induced cleavage of RPTP␣ in pri-mary cortical neurons did not proceed to completion. A similar finding was noted previously in the case of calpain-mediated cleavage of the STEP tyrosine phosphatase in vivo. STEP is cleaved by calpain in primary neuronal cells following glutamate stimulation (13) and in perinatal rat brain following hypoxia/ischemia (12), but in both cases cleavage is rather limited and does not deplete the full-length STEP precursor molecules. This could indicate that cleavage is regulated by molecular mechanisms in addition to activation of calpain. In the case of RPTP␣, it should be noted that cleavage occurs in close proximity to the wedge domain of the phosphatase (Fig.  2). This region participates in significant intermolecular interactions in the course of RPTP␣ dimerization (40), possibly impeding access of calpain to its site of action. Studies in transfected 293 cells have indicated that dimerization of RPTP␣ is quite prevalent, and that most RPTP␣ molecules present at the surface of these cells are found in dimers (40). Alternatively, the stimuli used in this study and in the studies of STEP might not have been sufficient to achieve full cleavage.
Of note, addition of 200 mM glutamate together with CaCl 2 to the cells examined here did not induce cleavage of RPTP␣ beyond that obtained with calcium alone. Homodimerization of RPTP␣ (40) and of PTP⑀ 3 could also explain the finding that, although p66 PTP␣ and p65 PTP⑀ are inherently cytoplasmic molecules, both can be found in part at the cell membrane when expressed together with their fulllength RPTP␣ or PTP⑀ precursors ( Fig. 7; Ref. 16). Dimerization of RPTP␣ molecules is believed to be mediated by interactions throughout the entire RPTP␣ molecule, which involve the extracellular and transmembranal domains, the juxtamembranal wedge domain of the membrane-proximal catalytic domain, and the membrane-distal catalytic domain (40  Tyrosine-phosphorylated proteins were immunoprecipitated (IP) from cell lysates, blotted, and probed with antibodies against Kv2.1 or Src, and were then analyzed with a scanning densitometer. Bar diagram shows relative amounts of phospho-Kv2.1, normalized to Kv2.1 expression levels. Note the strong effect of RPTP␣ on Kv phosphorylation, despite its much weaker expression compared with p66 PTP␣. Blots and diagram are from an experiment representative of three performed. WB, Western blot. age of RPTP␣ within the juxtamembrane domain, which removes the transmembranal and extracellular domains of RPTP␣ but does not remove the wedge domain or the remainder of the catalytic domains, should therefore not necessarily abolish these interactions. We believe these interactions would withstand the process of cell fractionation, as PTP⑀ homodimers can readily be detected following immunoprecipitation. 3 It remains to be determined whether membrane-associated p66 PTP␣ and p65 PTP⑀ are molecules that have remained bound to their original uncleaved dimerization partner, or have been recruited to intact membrane-associated RPTP␣ or PTP⑀ monomers.
Proteolytic cleavage and subcellular re-distribution of PTP⑀ and PTP␣ is expected to affect the physiological functions of both PTPases in a significant and irreversible manner. Two possible outcomes of this process that are not mutually exclusive are loss of function toward membrane-associated substrates and gain of function toward substrates located in the cytosol. As very little is known concerning the cytosolic functions of both PTPases, we chose to focus here on the loss-offunction consequences of cleavage. Indeed, both p65 PTP⑀ and p66 PTP␣ are severely limited in their ability to perform physiologically relevant roles, which depend on their being present at the cell membrane. Reduced ability of p66 PTP␣ to dephosphorylate Src at Tyr-529 is of particular significance, as regulation of Src phosphorylation and activity is perhaps the best characterized role of RPTP␣ to date. Dephosphorylation by RPTP␣ at their C-terminal negative regulatory tyrosine activates Src and the related Fyn kinase (28 -32). This, in turn, causes several key physiological outcomes, including cellular transformation (32) and modulation of cell adhesion and spreading (31,33). Altered phosphorylation and activation of Src by PTP␣ has been correlated with physiological consequences in RPTP␣-deficient mice (30,31), attesting to the relevance of our findings.
p66 PTP␣ is also significantly less able than RPTP␣ to reduce phosphorylation of the Kv2.1 voltage-gated potassium channel, an observation similar to that made previously in the case of p65 PTP⑀ and p67 PTP⑀ (16). Dephosphorylation of Kv2.1 by PTP⑀ is of clear physiological importance in vivo, as it affects Kv2.1 channel activity in Schwann cells; it is also correlated with transient severe hypomyelination of sciatic nerves of PTP⑀-deficient mice (25). This study suggests that RPTP␣ might also affect Kv2.1 channel activity by altering its phosphorylation state, although this remains to be verified experimentally. Membrane association plays an important role in mediating RPTP␣ function in other systems as well. RPTP␣ can down-regulate insulin receptor signaling in baby hamster kidney cells (19,20), possibly by dephosphorylating the ␤ subunit of the receptor (34). RPTP␣-induced inhibition of insulin receptor function has been shown to lead to decreased insulinstimulated prolactin gene expression (53). Membrane association is crucial here as well, as removing the transmembranal and extracellular domains of tm-PTP⑀ or RPTP␣ abolishes their ability influence insulin receptor signaling in this system (19).
Several points argue in favor of the possibility that cleavage and translocation allow PTP⑀ and PTP␣ access to potentially new substrates in the cytosol. First, the cleaved products are ubiquitously expressed and stable, and studies have shown that p65 PTP⑀ molecules, as well as PTP␣ molecules devoid of their extracellular and membrane-spanning domains, are catalytically active (16,54). 4 Second, a cytosolic gain of function could affect physiological processes even without full cleavage of tm-PTP⑀ or RPTP␣. However, direct testing of this possibility requires additional information concerning the cytosolic functions on both PTPases. A third potential consequence of cleavage is physical and irreversible separation between the catalytic domains of RPTP␣ and tm-PTP⑀ and the extracellular domains of these molecules. Recent studies have suggested that interactions between the extracellular domain of RPTP␣ and extracellular molecules exist (43,44). Their physiological consequences, however, remain to be determined before this possibility can be adequately addressed.
A 75-kDa processed form of RPTP␣, which is induced in NIH3T3 cells upon treatment with pervanadate, has been described recently (55). In contrast to this form, p66 PTP␣ is detected in a constitutive manner in most tissues and cell lines where RPTP␣ is detected, and neither it nor p65 PTP⑀ is induced by pervanadate treatment (results not shown). Furthermore, our data indicate that p66 PTP␣ and p65 PTP⑀ are the products of cleavage that occurs in the cytoplasmic juxtamembrane region of RPTP␣, rather than on the outside of cells as has been suggested for the 75-kDa protein. The possible relationship between this protein and p66 PTP␣ therefore remains to be established.
This study highlights several new points of similarity and difference between PTP␣ and PTP⑀. Similarities extend to the existence of proteolytically processed forms of both phosphatases, to the mechanisms by which they are produced, and to the effects cleavage has on reducing the ability of both phosphatases to act on membrane-associated substrates. However, similarities between PTP␣ and PTP⑀ are not absolute, as no forms of PTP␣ that are analogous to full-length cyt-PTP⑀ or to the internal initiation product p67 PTP⑀ have been found to date despite intensive searches. In all, expression of PTP⑀ and PTP␣, which together now include six distinct protein forms, is subject to complex regulation at the levels of transcription, translation, and post-translational processing. The concept that altering subcellular localization is a major factor in regulating the functions of non-membranal PTPases is well established (2). The existence of processed forms of PTP␣ and PTP⑀ and their altered physiological properties as compared with their full-length precursors indicates that this principle operates among membrane-bound PTPases as well.